r - 如何将位置分配为聚类质心并运行聚类以找到离每个质心最近的点？

Question

使用质心坐标作为质心，并根据最近的位置对它们周围的其余位置进行聚类。

Centroid <- data.frame (longitude  =c( -1.482880, -1.485735),
                            latitude = c( 54.89935, 54.89935),
                            ID = c(1,2,3,4,5))

Locations <- data.frame(longitude = c(-1.482156, -1.482318, -1.482129, -1.484275, -1.485866), 
                           latitude= c(54.90083, 54.90078, 54.90077, 54.90011, 54.89936),
                           ID = c(A,B,C,D,E))

标签： rgeospatial

Answer 1

使用 sf 包查找离位置最近的质心相对容易。您需要将数据框转换为 sf 对象，如下所示。

library(tidyverse)
library(sf)
#> Linking to GEOS 3.6.2, GDAL 2.2.3, PROJ 4.9.3


centroid <- data.frame (longitude  =c( -1.482880, -1.485735),
                        latitude = c( 54.89935, 54.89935),
                        ID = c(1,2))

locations <- data.frame(longitude = c(-1.482156, -1.482318, -1.482129, -1.484275, -1.485866), 
                        latitude= c(54.90083, 54.90078, 54.90077, 54.90011, 54.89936),
                        ID = c(LETTERS[1:5]))

# 4326 is common, but a poor choice for accurate distances. you should choose a better crs.
cen_sf <- st_as_sf(centroid, coords = c('longitude', 'latitude')) %>% st_set_crs(4326)
loc_sf <- st_as_sf(locations, coords = c('longitude', 'latitude')) %>% st_set_crs(4326)

st_nearest_feature(loc_sf, cen_sf)
#> although coordinates are longitude/latitude, st_nearest_feature assumes that they are planar
#> [1] 1 1 1 1 2
st_distance(loc_sf, cen_sf)
#> Units: [m]
#>          [,1]      [,2]
#> [1,] 171.1779 282.59946
#> [2,] 163.2218 270.91505
#> [3,] 165.2558 280.18680
#> [4,] 123.1550 126.21773
#> [5,] 191.5677   8.47761

^{由reprex 包于 2020-01-16 创建(v0.3.0)}